How you show up
Your feelings, traits and habits, read from your own words.
Private conversation intelligence
MOSAIC reads the conversations you choose and turns months or years of messages into a private portrait of you, the people who matter to you, and the dynamics between you.
What MOSAIC does
Pick the conversations you want to understand. MOSAIC reads them and turns them into private reports about you, the people you talk to, and how you communicate together.
01 · Your messages
02 · The analysis
03 · The result
The idea
Every lens reads a relationship from three sides: how you show up, how they show up, and what the two of you create together.
Your feelings, traits and habits, read from your own words.
The same read of the other person, side by side with yours.
The patterns neither of you makes alone: who leads, who adapts, how you clash and how you repair.
Across people · Constellation
This is the Constellation report. Put several relationships together and MOSAIC shows what stays the same with everyone, and what only certain people bring out in you.
About the Constellation report →Example data.
Five lenses
Personality, narcissism, emotional regulation & self-concept, romantic relationships, and spiritual development & worldview. Each lens reads the same conversations differently. Use one, or combine them for the fullest picture.
01 / 05
How you're both wired, and where you meet.
02 / 05
See how self-focus and control show up, on both sides.
03 / 05
Emotional regulation & self‑concept
See how feelings move, and how you speak to yourself.
04 / 05
The story of the two of you.
05 / 05
Spiritual development & worldview
Where you are each seeing from, and how you grow together.
Two ways to see it
You and one person
A full profile of you, a full profile of them, and how the two of you interact. Two complete portraits and the dynamic between them, from one conversation.
You across five or ten people
Your conversations with five or ten people, read together. It separates what belongs to one relationship from the patterns you bring to all of them.
Why MOSAIC
ChatGPT, Claude and Gemini are brilliant generalists. But paste in a chat and you get one model’s impression of whatever it took in. Here’s what that misses.
01
Bury one important message in a long prompt and models often miss it. Researchers call it the needle-in-a-haystack problem, and in a relationship that one message can explain everything else. The longer the chat, the more slips through, well before the model reaches its limit. MOSAIC sends hundreds, often thousands, of small focused prompts instead, so every message gets a close read.
SourcesModarressi et al., 2025·Chroma, 2025·Song et al., 2025
Pasted into a chatbot
One long prompt
Prompts sent1
One pass. Pieces get lost.
MOSAIC
Split into sections
Each segment, read for each trait
Prompts sent1,248
Hundreds of focused reads, every message covered.
02
A chatbot is one model reading, deciding and writing in a single pass, with nothing checking its work, and models are poor at catching their own mistakes. Researchers call it the self-correction blind spot. MOSAIC adds message statistics and validated word lexicons, and an independent re-check reads the evidence again before any finding is kept.
Pasted into a chatbot
Not checked
Not checked
Not checked
One model, one pass, one answer.
MOSAIC
Kept
Kept
Dropped
Statistics, lexicons and models, then a re-check.
03
Ask a chatbot what’s going on and it reaches for whatever words come to mind: distant, mixed signals, clearly cares. They sound right, but they aren’t defined, so there’s nothing to measure them against. Psychologists call this the jingle-jangle problem, where one word means different things and different words mean the same thing. MOSAIC reads for hundreds of narrow traits defined in published psychology research, like attachment, the Big Five and the pursue–withdraw pattern, and scores each one on its own evidence.
SourcesWulff & Mata, 2026·Hazan & Shaver, 1987·Christensen & Heavey, 1990·Driver & Gottman, 2004·Paunonen & Ashton, 2001·Zhu et al., 2025
Pasted into a chatbot
“mixed signals”
Words that sound right.
MOSAIC
AttachmentHazan & Shaver, 1987
Turning toward each otherDriver & Gottman, 2004
Big FiveJohn & Srivastava, 1999
J. Pers. Soc. Psychol. · 1990
Gender and social structure in the demand/withdraw pattern of marital conflict
Christensen & Heavey
Pursue–withdraw
Defined traits, with their sources.
04
A chatbot sounds just as sure when it’s wrong, and it can misquote you or invent details outright. Researchers call it faithfulness hallucination. In MOSAIC, code checks every quote against your real messages, borderline findings get a second read, and the confidence score comes from what survives.
SourcesHuang et al., 2025·Kalai et al., 2025·Ansari, 2026·Wu et al., 2025·Cash et al., 2025
Pasted into a chatbot
Stated confidence
“I’m very confident.”
Confident, with nothing to check.
MOSAIC
Computed confidence
1 message against78%
Quotes checked, linked and scored.
See all the reasons MOSAIC is better.
Learn about every way MOSAIC beats a chatbot, from how it reads your messages to how it keeps them private.
Learn morePrivacy
Your message database never leaves your Mac or phone. MOSAIC reads it where it lives.
Phone numbers and email addresses are replaced with anonymous IDs before anything is sent.
Every report is tied to your account and can only be opened by you, unless you choose to share it.
Platforms
iMessage · Contacts
Reads your Messages history and Contacts on your computer, straight from a Mac or from your iPhone’s backup on a PC, so you can analyze years of iMessage conversations.
WhatsApp export
Import a WhatsApp chat export in a few taps and read your reports wherever you are.
Get started
Choose a chat that matters to you and see what months or years of messages have been saying all along. Or bring in five or ten, and see the patterns that are yours everywhere.